Rose Forms as determined by their Cy to logical Behaviour . 185 
Since our conceptions of the status of the British roses, derived from the work 
outlined above and otherwise, differ profoundly from those held up to the 
present by most, but certainly not all, rose specialists, the ground is not 
prepared for any definite utilization of the observed chromosome numbers. 
Notwithstanding this one cannot help being struck by the fact that all 
of the supersection Caninae, minus the Villosae, are pentaploid forms, which 
confirms an alliance between the more widely separated R. tomentosa 
and R. foetida groups on the one hand, and the Eucaninae on the other, 
suggested by many similarities more apparent on the growing bushes. In 
much the same manner the agreement in tetraploidy between the obviously 
related R. mollis, R. omissa, and R. Sherardi forms serves to strengthen 
the case for our previous removal of the whole of these groups to the 
Villosae. Hence we see .that the old classification of R. omissa and R. 
* Sherardi as tomentosa segregates, upon no more secure basis than the 
non-persistence of their sepals, fails both when judged from a cytological 
standpoint and from their undoubted approximation in general charac- 
teristics when one examines them otherwise than as dried fragments. 
Summary. 
1. The fundamental chromosome number in Rosa is seven. 
3 . Among the roses examined were diploid, tetraploid, pentaploid, and 
hexaploid forms. 
3. No diploid form was found to be abnormal during meiosis ; nor was 
any member of the pimpinellifolia section among the tetraploid. 
4. The remainder of the tetraploids, the whole of the pentaploids and 
hexaploids, showed a partial reduction involving fourteen or twenty-eight 
chromosomes. 
5. In these partially reduced forms the heterotype anaphase occurred 
in two steps, one involving the chromosomes to which we have just referred, 
and the second, and later, taking place when the univalents split. 
6. In many cases these split univalents failed to reach the poles 
and formed supernumerary micronuclei in the interkinesis between the 
heterotype and homotype divisions. 
7. The features mentioned in 5 and 6 are exaggerated in the homotype 
division, so that, for the most part, the major nuclei in the tetrad are 
endowed with seven chromosomes in the pentaploids and abnormal tetra- 
ploids. 
8. When the remainder of the chromosomes keep in two groups, eight 
major nuclei may arise, thus yielding an octad rather than a tetrad. 
9. Multinucleate pollen grains are quite common in the anomalous 
forms of the genus Rosa, but when these are derived chiefly from micro- 
nuclei the pollen grains collapse. 
